To be clear: the reason why I find this so surprising is because I was arriving here from the 'cash transfers in the developing world' literature, where giving people money absolutely, comprehensively solves many of their problems. They eat more. Their kids are less likely to die. They improve the quality of their housing. They take needed medications more. I didn't have a first principles conviction that cash helps people - I'd watched study after study where cash helps people.
Even in the US, I think you much overstate the case that the welfare state should be expected to do approximately nothing. There are well designed studies of the expansion of Medicaid that find reductions in all cause mortality. SNAP's rollout and expansion genuinely reduced instances of people going hungry. I think what you can't expect in the US is compounding benefits (predictably or at any scale), where having more resources allows someone to take a better job and fix their health and then be sustainedly much better off than without the intervention; we're a 'efficient market' in that the people who can have this trajectory mostly in fact do have this trajectory.
On a bigger picture level, I think that any given anti-poverty intervention probably doesn't work - but I think this in the same spirit that I think any given startup probably won't work. A lot of shots are worth taking because the expected value of success is extraordinary, even if the expected result is not much.
Is there more return in replying on Substack or X? I think probably the former...
Vis-a-vis the African studies, I tend to think that "works in Africa" is the social policy equivalent of "works in vitro" or "works in mice" in the bio literature. That is, it shouldn't shift our priors very much at all, because the margin on which the policy is operating is so different. When your marginal dollar is a decision between starvation or not, it's way more likely to get measurable impact than when you already live in a robust social safety net that mostly prevents those outcomes.
But that conditional matters a lot for the marginal social policy intervention! It's hard to actually starve to death or die from otherwise curable illness in the United States, relative to both less developed nations and to our past. So the easy to access pathways for improvement are cut off. Yet many kinds of pathology durably persist! The fact that crime, OD deaths, etc. (yes, sure, I am primed to think about these) do not seem to be related to wealth should give us pause about assuming transfers do much to address these issues. (And this was the context in which Rossi was writing — not when you're at <1K GDP/capita, but when you're at developed world GDP/capita.)
Which gets to your bigger point. The case for the unicorn in startup world is that there's some very large, hard-to-identify return. But U.S. social welfare is ... pretty good in absolute terms? I'm not in favor of getting rid of UI or SSDI. But by the same token, because we have them, it's not obvious to me that we should be doing more experimentation in social policy to try to address poverty per se. Especially not when the most efficient way to lift people out of poverty is just growth!
(btw, on that Medicaid mortality study, I think the findings are slightly more complicated than the initial framing, see Robert Ver Bruggen's write up for us: https://www.city-journal.org/article/medicaid-reform-expansion-republicans-health-insurance. But also, I kind of suspect given the age band of the sample and the prior lit on what Medicaid does, that most of the effect is in SUD treatment leading to fewer OD deaths. Which isn't nothing! But is not a comprehensive effect.)
First, I love this and I hope both you and Ms. Piper expand in long form. You've hit on a really important difference here.
Second, though I mostly agree with your point about "what margin are we even on here guys?" I think you do go wrong in an important way in your fourth paragraph.
The cash transfer studies in America have come at objectively a minor cost but provided extremely useful information. That seems to be Ms. Piper's point that experimentation is useful. This type of experiment, solidly based on a hypothesis, can tell us a lot about what tweaks at our current margin we should make.
I really lose you when you state it's "not obvious [. . .] that we should be doing more experimentation in social policy to address poverty per se." The phrases "not obvious" and "doing more experimentation" are both doing a lot of lifting...how much of this experimentation are we doing? Was this particular experiment a useful contribution? Are we doing the 'right amount' of experimentation such that we are perfectly at the right margin with respect to the amount of experimentation?
(I'd turn this around and say my prior is the opposite: it's not obvious we're doing even close to enough experimentation in social policy to address poverty. Even from a conservative social view, this statement must be correct, no? Solid evidence Social Security cuts, at the margin, are not harmful would be amazing to conservatives?)
I read you to say something like: 'even the relatively minor cost of the study is not worth bearing because we already know the answer.' But that cannot be right a priori, without understanding with specificity how much this cost and how much was born by the public.
After all, if it's private money on the line, almost by definition, it was a worthwhile endeavor.
Just wrote this in another comment, but I think defenders of the social safety net could do more to distinguish between marginal and non-marginal effects. The marginal effects of additional transfers are zero, but the effects of the pre-existing safety net may be positive, just like the effects of cash transfers in developing countries.
The main distinction between cash transfers in the USA and Africa is probably the one Lehman emphasizes about the impact of a marginal dollar being substantial for someone at the brink of starvation, but undetectable, otherwise.
But I think there's another, intuitive, distinction - The Sort.
I suspect that while you'd support unconditional cash transfers in Africa, you wouldn't personally engage in them for the only Americans who might have a comparable standard of living - the homeless.
I'd guess that you'd offer a homeless American food, rather than money. Why would that be? Because you'd intuitively realize that the money would very likely go towards drugs, rather than anything productive.
If those guesses are correct, presumably the reasoning would be a recognition of The Sort in the USA. Americans who are homeless are extremely negatively selected as far as traits like impulse-control and conscientiousness, while being born in Sub-Saharan Africa doesn't have the same implication.
Given that the relatively free market in the West allows for extensive social stratification based on traits, The Sort isn't limited to the homeless, so the poor, in general, will be negatively selected (and that's increasingly the case, as stratification increases).
That's also why intuition like:
>If you gave me an extra 10% of my income to spend, it would absolutely make me better off
breaks down. You're extremely *positively* selected, as opposed to the recipients who are extremely *negatively* selected.
Given the increasing Sort, we should expect interventions to become even less effective in the future, and for skepticism like Rossi's to become even more ironclad.
But why ignore the RCTs? Even quasi experimental designs that seem reasonable are vulnerable to the problem of file drawer bias/publication bias where you’re not seeing all of the analysis that was actually done.
"I think that any given anti-poverty intervention probably doesn't work - but I think this in the same spirit that I think any given startup probably won't work."
This is just completely and totally wrong. We know exactly how to solve poverty for any given population and it is direct cash transfer. We know this because without Social Security, the largest direct cash transfer program we have, 22 million more people would have been in poverty in 2024.
This is pretty easy to see. A 98 year old without Social Security (assuming they do not own income generating wealth) will be in poverty. Add in Social Security, and they are no longer in poverty. They may have severe mental or physical health issues. They may be homeless. They may have poor nutritional habits. Those are not poverty.
The same would be true of an 8 year old, however, our society doesn't have Social Security for children.
Both of you would have much cleaner analysis if you defined poverty as what it is: a lack of money. Then you could look at the populations that would be in poverty without non-labor direct cash transfers, and saw that direct cash transfers work to get people out of poverty.
It may or may not have other desirable outcomes that you'd like to see (improved mental or physical health, better employment etc.), but you are confusing those outcomes with poverty.
I agree with a lot in this post, but it could do more to acknowledge -- and draw inferences from -- the clear success of cash transfers in developing countries.
To me this suggests a distinction between the effects of marginal transfers above and beyond the existing safety net, and the safety net itself.
It is quite possible that the existing safety net provides many of the same real benefits as cash transfers in developing countries, as they are both riding atop little to no income.
An up-to-date version of Rossi’s conundrum is Megan Stevenson’s CAUSE, EFFECT, AND THE STRUCTURE OF THE SOCIAL WORLD (Boston Univ. Law Review 2002). Focusing on Charles’ favorite topic criminal justice reform she argues:
When it comes to the type of limited-scope interventions that lend themselves to high-quality evaluation, social change is hard to engineer. Stabilizing forces push people back toward the path they would have been on absent the intervention. … [C]ausal processes are complex and context dependent, meaning that a success achieved in one setting may not port well to another. This has a variety of implications. It suggests that a dominant perspective on social change—one that forms a pervasive background for academic research and policymaking —is at least partially a myth.
Some of the push-back regarding this line of argument has focused on its implications for evaluation, which constituted the bulk of Stevenson’s discussion.(1) She sees the social world as too complex, resilient and multi-causal to be influenced by the kinds of interventions that can be funded and launched, and that researchers can evaluate using the tools at hand. She would conclude that the anti-poverty projects Charles describes fall short of reforms that ever had a prayer of changing their targets. Stevenson calls instead for ‘big push’ (my term) interventions that tackle constraints on changing people’s behavior which are “deep” and “structural” but (and this is also my reading) probably not allowable by the extant political and economic order.
(1) See: Brandon del Pozo, Steven Belenko, Faye S. Taxman, Robin S. Engel, Jerry Ratcliffe, Ian Adams and Alex R. Piquero. Then a miracle occurs: cause, effect, and the heterogeneity of criminal justice research. Journal of Experimental Criminology, https://doi.org/10.1007/s11292-024-09636-7 (online-not yet in print).
I went into the new substack understanding it as an attempt to Trojan-horse better ideas into leftist discourse, most of the article is actually just signaling in-group membership, a spoonful of sugar with the medicine. You can't just tell them "the poor are poor for a reason", you will be ignored. Instead you must judo their existing ideas into data driven policy.
Infrastructure, effective schools, medical advancements, dental care, and emergency (health) insurance would be my best guesses at which kinds of "transfer" investments most effectively redistribute the benefits of growth
How strong is the evidence for effective schools having a long term impact? My understanding is that all interventions in improving schools have a null effect.
i think the elephant in the room is that american social policy is backward. academically backward and insular. the big european advances are ignored, as if esping-andersen, korpi and palme, or even myrdal’s work in the united states had never existed. maybe they read them, but nothing shows in the system or the debate. europeans understood long ago the limits of targeting, but those insights never touched american scholars, who stay busy with endless policy evaluations and ad-hoc patches while missing the larger picture.
what’s even stranger is that american academia is still busy wrapping its head around targeting, or the tired debate on in-cash versus in-kind welfare. of course welfare needs a large share of in-kind services... that’s... obvious. but it’s as if nobody in the us has read the decades of work produced on the other side of the atlantic. the debate keeps circling back to basic questions that european social science treated as settled long ago.
european sociologist are architects, with vision and design. american social policy is plumbing, scrambling to fix leaks without any idea where the whole structure is going. it is baffling. anyone serious about studying social policy should go to denmark or sweden instead of wasting time in the united states, recycling the same blind evaluations of silly and dumb "programs".
edit: i think we should also talk about the various "experimental approaches to alleviating global poverty" that have ravaged development studies for far too long in america. it is an analogous situation to what i said above, and i think Angus Deaton was right on target in pointing out its limits.
Look at the authors I cited above to start. Gøsta Esping-Andersen: In The Three Worlds of Welfare Capitalism, the opening chapters explicitly criticize the social administration tradition for being typological, atheoretical. And Korpi criticized mainstream welfare research for treating welfare policy as technical and neutral rather than as part of class conflict. In their subsequent work, they analyzed welfare not by doing the same retarded policy evaluations that this post is debating. The Korpi and Palme paradox is that the more you target, and the more you use cash benefits, the more poverty is reiterated instead of challenged. In order to fight poverty, you need universal services (such as kindergartens, training, eldercare, etc) and not only some money transfers. These intuitions have been researched empirically in every possible way, suggesting a better way to build a comprehensive welfare state and social policies. If you limit yourself to the study of the effects of single programs, you miss the forest for the trees.
I'd also rephrase it, as it presents political participation as inherently positive. In reality, voting is zero sum, so the net effect is basically 0. Voters mostly help inasmuch as they're better at voting than average, and by definition, not everyone is better than average.
In fact, marginal voters on fence about political participation, are generally less informed than average, and their participation is probably a net negative.
Improved political participation would be them staying home and watching football.
a) are people in situations X outcomes improved by providing them with additional resources?
b) are people in situations X whose outcomes improved ARE improved by giving them by providing them with additional resources improved more by giving them cash rather than in kind resources of the same value minus administrative costs
Is seems the studies cited do not address these issued clearly, although they are more relevant to a than b.
It would be surprising if there are not some X for whihc the answer to a) is yes and others no.
It would be surprising if there are not some people for whom cash
is better and other for whom in kind id better.
I think Medicaid is better than enough cash to buy a health insurance policy
I think a cash child allowance is better than public day care costing the same amount.
But the studies are not relevant to testing those hypotheses
Many thanks for this post. I found it so interesting that I also read the 2024 Fusion article linked in footnote 2. Here is a quote from the latter piece: "Above all, we can learn that social dysfunction is not merely about deprivation. Just giving people money can do very little to change their behavior for the better." I am not at all well-read in the literature that this Substack post, the Substack post by Ms. Piper, and the Fusion article are a part of. Maybe that it is why if A had told me that B claims (conjectures) that there is a stable cause-and-effect relationship running from financial resources to indicators of social dysfunction (as implied by the first sentence cited) I would have considered that a strawman argument advanced by A by discredit B. It is a patently ludicrous claim bordering on magical thinking that no one with a bit of life experience outside academia can take seriously.
Can we agree that the "social dysfunction" being referred to is the direct consequence of a "set of dysfunctional behaviors" in the sense that these behaviors are, at a minimum, a necessary condition for this "social dysfunction" to arise? Then there is no alternative to changing (stopping) these behaviors if we wish the "social dysfunction" to end. In concrete terms, we are talking about changes such as doing one's home-work even though it would be more enjoyable to go out and play, not trying drugs and not engaging in criminal activity even though one's peers may consider these activities cool, having children only when certain preconditions are met, not missing medical appointments and so on. These changes would have to be made not once or for a few weeks, but consistently for a period of many years. That anyone would think these behavioral changes may be effected by simply giving money to the individuals in question is a belief so naive and simplistic that it is almost beyond my imagination.
Having said that, I am the first to admit that an alternative prescription that I almost reflexively think of myself, namely getting individuals' incentives right, is a similarly simplistic recipe of the "black box" type. Some careful thinking is called for. Someone, somewhere - an anthropologist perhaps? - must have developed a more precise understanding of what the roots of the problem(s) are which, in turn, may give rise to more intelligent and thus more effective approaches.
I hope to read more about that topic both here on Substack and in City Journal (and elsewhere). As far as I recall the three pieces, I did not come across any hints as to what "works" or might "work". Ms. Piper asserts that "Winning the war on poverty will require more than just transfers, it will require building and improving institutions that provide education, health care and housing.", but provides no evidence supporting this somewhat nebulous claim. Could it not be the case that the most helpful interventions turn out to be entirely nonintuitive?
Those successful early trials were the reason that people tried larger, well designed RCTs that did not reproduce the benefits those early trials found.
Yes. Going back even further, Gray Areas is another example of a pilot program (and not a very successful one at that!) being scaled up to LBJ levels to dismal effect:
I want to resist the hardass analysis of commenter Keith but he does have a point. As long as foundations are involved in these efforts, and their leaders continue to reinforce each other in their echo chambers of idealistic nonsense, the impact of trenchant analyses like this one will be net zero.
Thanks for the thoughtful response.
To be clear: the reason why I find this so surprising is because I was arriving here from the 'cash transfers in the developing world' literature, where giving people money absolutely, comprehensively solves many of their problems. They eat more. Their kids are less likely to die. They improve the quality of their housing. They take needed medications more. I didn't have a first principles conviction that cash helps people - I'd watched study after study where cash helps people.
Even in the US, I think you much overstate the case that the welfare state should be expected to do approximately nothing. There are well designed studies of the expansion of Medicaid that find reductions in all cause mortality. SNAP's rollout and expansion genuinely reduced instances of people going hungry. I think what you can't expect in the US is compounding benefits (predictably or at any scale), where having more resources allows someone to take a better job and fix their health and then be sustainedly much better off than without the intervention; we're a 'efficient market' in that the people who can have this trajectory mostly in fact do have this trajectory.
On a bigger picture level, I think that any given anti-poverty intervention probably doesn't work - but I think this in the same spirit that I think any given startup probably won't work. A lot of shots are worth taking because the expected value of success is extraordinary, even if the expected result is not much.
Is there more return in replying on Substack or X? I think probably the former...
Vis-a-vis the African studies, I tend to think that "works in Africa" is the social policy equivalent of "works in vitro" or "works in mice" in the bio literature. That is, it shouldn't shift our priors very much at all, because the margin on which the policy is operating is so different. When your marginal dollar is a decision between starvation or not, it's way more likely to get measurable impact than when you already live in a robust social safety net that mostly prevents those outcomes.
But that conditional matters a lot for the marginal social policy intervention! It's hard to actually starve to death or die from otherwise curable illness in the United States, relative to both less developed nations and to our past. So the easy to access pathways for improvement are cut off. Yet many kinds of pathology durably persist! The fact that crime, OD deaths, etc. (yes, sure, I am primed to think about these) do not seem to be related to wealth should give us pause about assuming transfers do much to address these issues. (And this was the context in which Rossi was writing — not when you're at <1K GDP/capita, but when you're at developed world GDP/capita.)
Which gets to your bigger point. The case for the unicorn in startup world is that there's some very large, hard-to-identify return. But U.S. social welfare is ... pretty good in absolute terms? I'm not in favor of getting rid of UI or SSDI. But by the same token, because we have them, it's not obvious to me that we should be doing more experimentation in social policy to try to address poverty per se. Especially not when the most efficient way to lift people out of poverty is just growth!
(btw, on that Medicaid mortality study, I think the findings are slightly more complicated than the initial framing, see Robert Ver Bruggen's write up for us: https://www.city-journal.org/article/medicaid-reform-expansion-republicans-health-insurance. But also, I kind of suspect given the age band of the sample and the prior lit on what Medicaid does, that most of the effect is in SUD treatment leading to fewer OD deaths. Which isn't nothing! But is not a comprehensive effect.)
First, I love this and I hope both you and Ms. Piper expand in long form. You've hit on a really important difference here.
Second, though I mostly agree with your point about "what margin are we even on here guys?" I think you do go wrong in an important way in your fourth paragraph.
The cash transfer studies in America have come at objectively a minor cost but provided extremely useful information. That seems to be Ms. Piper's point that experimentation is useful. This type of experiment, solidly based on a hypothesis, can tell us a lot about what tweaks at our current margin we should make.
I really lose you when you state it's "not obvious [. . .] that we should be doing more experimentation in social policy to address poverty per se." The phrases "not obvious" and "doing more experimentation" are both doing a lot of lifting...how much of this experimentation are we doing? Was this particular experiment a useful contribution? Are we doing the 'right amount' of experimentation such that we are perfectly at the right margin with respect to the amount of experimentation?
(I'd turn this around and say my prior is the opposite: it's not obvious we're doing even close to enough experimentation in social policy to address poverty. Even from a conservative social view, this statement must be correct, no? Solid evidence Social Security cuts, at the margin, are not harmful would be amazing to conservatives?)
I read you to say something like: 'even the relatively minor cost of the study is not worth bearing because we already know the answer.' But that cannot be right a priori, without understanding with specificity how much this cost and how much was born by the public.
After all, if it's private money on the line, almost by definition, it was a worthwhile endeavor.
Just wrote this in another comment, but I think defenders of the social safety net could do more to distinguish between marginal and non-marginal effects. The marginal effects of additional transfers are zero, but the effects of the pre-existing safety net may be positive, just like the effects of cash transfers in developing countries.
The main distinction between cash transfers in the USA and Africa is probably the one Lehman emphasizes about the impact of a marginal dollar being substantial for someone at the brink of starvation, but undetectable, otherwise.
But I think there's another, intuitive, distinction - The Sort.
I suspect that while you'd support unconditional cash transfers in Africa, you wouldn't personally engage in them for the only Americans who might have a comparable standard of living - the homeless.
I'd guess that you'd offer a homeless American food, rather than money. Why would that be? Because you'd intuitively realize that the money would very likely go towards drugs, rather than anything productive.
If those guesses are correct, presumably the reasoning would be a recognition of The Sort in the USA. Americans who are homeless are extremely negatively selected as far as traits like impulse-control and conscientiousness, while being born in Sub-Saharan Africa doesn't have the same implication.
Given that the relatively free market in the West allows for extensive social stratification based on traits, The Sort isn't limited to the homeless, so the poor, in general, will be negatively selected (and that's increasingly the case, as stratification increases).
That's also why intuition like:
>If you gave me an extra 10% of my income to spend, it would absolutely make me better off
breaks down. You're extremely *positively* selected, as opposed to the recipients who are extremely *negatively* selected.
Given the increasing Sort, we should expect interventions to become even less effective in the future, and for skepticism like Rossi's to become even more ironclad.
> There are well designed studies of the expansion of Medicaid that find reductions in all cause mortality.
Which ones? I'm more familiar with Robin Hanson on the lack of effects from RCTs on health subsidies https://www.overcomingbias.com/p/karnataka-hospital-insurance-experimenthtml
On Twitter I said I liked https://sciencedirect.com/science/article/abs/pii/S0167629619306228 and https://ajph.aphapublications.org/doi/10.2105/AJPH.2025.308045. Cremieux had some methodological complaints, some generic to any DiD and none that convincing to me.
But why ignore the RCTs? Even quasi experimental designs that seem reasonable are vulnerable to the problem of file drawer bias/publication bias where you’re not seeing all of the analysis that was actually done.
https://x.com/Jabaluck/status/1454293779804876802?s=19
A very good thread on the topic.
"I think that any given anti-poverty intervention probably doesn't work - but I think this in the same spirit that I think any given startup probably won't work."
This is just completely and totally wrong. We know exactly how to solve poverty for any given population and it is direct cash transfer. We know this because without Social Security, the largest direct cash transfer program we have, 22 million more people would have been in poverty in 2024.
This is pretty easy to see. A 98 year old without Social Security (assuming they do not own income generating wealth) will be in poverty. Add in Social Security, and they are no longer in poverty. They may have severe mental or physical health issues. They may be homeless. They may have poor nutritional habits. Those are not poverty.
The same would be true of an 8 year old, however, our society doesn't have Social Security for children.
Both of you would have much cleaner analysis if you defined poverty as what it is: a lack of money. Then you could look at the populations that would be in poverty without non-labor direct cash transfers, and saw that direct cash transfers work to get people out of poverty.
It may or may not have other desirable outcomes that you'd like to see (improved mental or physical health, better employment etc.), but you are confusing those outcomes with poverty.
I agree with a lot in this post, but it could do more to acknowledge -- and draw inferences from -- the clear success of cash transfers in developing countries.
To me this suggests a distinction between the effects of marginal transfers above and beyond the existing safety net, and the safety net itself.
It is quite possible that the existing safety net provides many of the same real benefits as cash transfers in developing countries, as they are both riding atop little to no income.
An up-to-date version of Rossi’s conundrum is Megan Stevenson’s CAUSE, EFFECT, AND THE STRUCTURE OF THE SOCIAL WORLD (Boston Univ. Law Review 2002). Focusing on Charles’ favorite topic criminal justice reform she argues:
When it comes to the type of limited-scope interventions that lend themselves to high-quality evaluation, social change is hard to engineer. Stabilizing forces push people back toward the path they would have been on absent the intervention. … [C]ausal processes are complex and context dependent, meaning that a success achieved in one setting may not port well to another. This has a variety of implications. It suggests that a dominant perspective on social change—one that forms a pervasive background for academic research and policymaking —is at least partially a myth.
Some of the push-back regarding this line of argument has focused on its implications for evaluation, which constituted the bulk of Stevenson’s discussion.(1) She sees the social world as too complex, resilient and multi-causal to be influenced by the kinds of interventions that can be funded and launched, and that researchers can evaluate using the tools at hand. She would conclude that the anti-poverty projects Charles describes fall short of reforms that ever had a prayer of changing their targets. Stevenson calls instead for ‘big push’ (my term) interventions that tackle constraints on changing people’s behavior which are “deep” and “structural” but (and this is also my reading) probably not allowable by the extant political and economic order.
(1) See: Brandon del Pozo, Steven Belenko, Faye S. Taxman, Robin S. Engel, Jerry Ratcliffe, Ian Adams and Alex R. Piquero. Then a miracle occurs: cause, effect, and the heterogeneity of criminal justice research. Journal of Experimental Criminology, https://doi.org/10.1007/s11292-024-09636-7 (online-not yet in print).
I went into the new substack understanding it as an attempt to Trojan-horse better ideas into leftist discourse, most of the article is actually just signaling in-group membership, a spoonful of sugar with the medicine. You can't just tell them "the poor are poor for a reason", you will be ignored. Instead you must judo their existing ideas into data driven policy.
Infrastructure, effective schools, medical advancements, dental care, and emergency (health) insurance would be my best guesses at which kinds of "transfer" investments most effectively redistribute the benefits of growth
How strong is the evidence for effective schools having a long term impact? My understanding is that all interventions in improving schools have a null effect.
i think the elephant in the room is that american social policy is backward. academically backward and insular. the big european advances are ignored, as if esping-andersen, korpi and palme, or even myrdal’s work in the united states had never existed. maybe they read them, but nothing shows in the system or the debate. europeans understood long ago the limits of targeting, but those insights never touched american scholars, who stay busy with endless policy evaluations and ad-hoc patches while missing the larger picture.
what’s even stranger is that american academia is still busy wrapping its head around targeting, or the tired debate on in-cash versus in-kind welfare. of course welfare needs a large share of in-kind services... that’s... obvious. but it’s as if nobody in the us has read the decades of work produced on the other side of the atlantic. the debate keeps circling back to basic questions that european social science treated as settled long ago.
european sociologist are architects, with vision and design. american social policy is plumbing, scrambling to fix leaks without any idea where the whole structure is going. it is baffling. anyone serious about studying social policy should go to denmark or sweden instead of wasting time in the united states, recycling the same blind evaluations of silly and dumb "programs".
edit: i think we should also talk about the various "experimental approaches to alleviating global poverty" that have ravaged development studies for far too long in america. it is an analogous situation to what i said above, and i think Angus Deaton was right on target in pointing out its limits.
Interesting. Can you link to these European sociologists' ideas or policy suggestions? What exactly is different?
Look at the authors I cited above to start. Gøsta Esping-Andersen: In The Three Worlds of Welfare Capitalism, the opening chapters explicitly criticize the social administration tradition for being typological, atheoretical. And Korpi criticized mainstream welfare research for treating welfare policy as technical and neutral rather than as part of class conflict. In their subsequent work, they analyzed welfare not by doing the same retarded policy evaluations that this post is debating. The Korpi and Palme paradox is that the more you target, and the more you use cash benefits, the more poverty is reiterated instead of challenged. In order to fight poverty, you need universal services (such as kindergartens, training, eldercare, etc) and not only some money transfers. These intuitions have been researched empirically in every possible way, suggesting a better way to build a comprehensive welfare state and social policies. If you limit yourself to the study of the effects of single programs, you miss the forest for the trees.
"rather, the point is that the implicit causal model of these experiments (less money → more problems) "
Indeed, as discussed in Smalls, Betha, Combs, and Jordan (1997), a better model actually is Mo(re) Money → Mo(re) Problems
> If you’re Murray you talk about biology
I don't think he actually talks that much about biology. Herrnstein was the scientist/psychologist of the pair.
"Which of these explanations is correct?"
Hint: its Murray
The link for "Their political participation doesn’t improve" doesn't work for me, but this one does: https://openresearch-web.files.svdcdn.com/production/assets/documents/Documentation/w33214.pdf.
I'd also rephrase it, as it presents political participation as inherently positive. In reality, voting is zero sum, so the net effect is basically 0. Voters mostly help inasmuch as they're better at voting than average, and by definition, not everyone is better than average.
In fact, marginal voters on fence about political participation, are generally less informed than average, and their participation is probably a net negative.
Improved political participation would be them staying home and watching football.
There are two different issues.
a) are people in situations X outcomes improved by providing them with additional resources?
b) are people in situations X whose outcomes improved ARE improved by giving them by providing them with additional resources improved more by giving them cash rather than in kind resources of the same value minus administrative costs
Is seems the studies cited do not address these issued clearly, although they are more relevant to a than b.
It would be surprising if there are not some X for whihc the answer to a) is yes and others no.
It would be surprising if there are not some people for whom cash
is better and other for whom in kind id better.
I think Medicaid is better than enough cash to buy a health insurance policy
I think a cash child allowance is better than public day care costing the same amount.
But the studies are not relevant to testing those hypotheses
(01 of 04)
Many thanks for this post. I found it so interesting that I also read the 2024 Fusion article linked in footnote 2. Here is a quote from the latter piece: "Above all, we can learn that social dysfunction is not merely about deprivation. Just giving people money can do very little to change their behavior for the better." I am not at all well-read in the literature that this Substack post, the Substack post by Ms. Piper, and the Fusion article are a part of. Maybe that it is why if A had told me that B claims (conjectures) that there is a stable cause-and-effect relationship running from financial resources to indicators of social dysfunction (as implied by the first sentence cited) I would have considered that a strawman argument advanced by A by discredit B. It is a patently ludicrous claim bordering on magical thinking that no one with a bit of life experience outside academia can take seriously.
(02 of 04)
Can we agree that the "social dysfunction" being referred to is the direct consequence of a "set of dysfunctional behaviors" in the sense that these behaviors are, at a minimum, a necessary condition for this "social dysfunction" to arise? Then there is no alternative to changing (stopping) these behaviors if we wish the "social dysfunction" to end. In concrete terms, we are talking about changes such as doing one's home-work even though it would be more enjoyable to go out and play, not trying drugs and not engaging in criminal activity even though one's peers may consider these activities cool, having children only when certain preconditions are met, not missing medical appointments and so on. These changes would have to be made not once or for a few weeks, but consistently for a period of many years. That anyone would think these behavioral changes may be effected by simply giving money to the individuals in question is a belief so naive and simplistic that it is almost beyond my imagination.
(03 of 04)
Having said that, I am the first to admit that an alternative prescription that I almost reflexively think of myself, namely getting individuals' incentives right, is a similarly simplistic recipe of the "black box" type. Some careful thinking is called for. Someone, somewhere - an anthropologist perhaps? - must have developed a more precise understanding of what the roots of the problem(s) are which, in turn, may give rise to more intelligent and thus more effective approaches.
(04 of 04)
I hope to read more about that topic both here on Substack and in City Journal (and elsewhere). As far as I recall the three pieces, I did not come across any hints as to what "works" or might "work". Ms. Piper asserts that "Winning the war on poverty will require more than just transfers, it will require building and improving institutions that provide education, health care and housing.", but provides no evidence supporting this somewhat nebulous claim. Could it not be the case that the most helpful interventions turn out to be entirely nonintuitive?
Thanks, I've correct.
corrected*. You can see where the copyediting problem came in...
If you use GPT to make your argument for you, you can also use GPT to make my argument against its argument for you.
A cool thing about my substack is that you don't get to come on it and insult me.
Those successful early trials were the reason that people tried larger, well designed RCTs that did not reproduce the benefits those early trials found.
Yes. Going back even further, Gray Areas is another example of a pilot program (and not a very successful one at that!) being scaled up to LBJ levels to dismal effect:
https://whorulesamerica.ucsc.edu/local/ford_foundation.html
I want to resist the hardass analysis of commenter Keith but he does have a point. As long as foundations are involved in these efforts, and their leaders continue to reinforce each other in their echo chambers of idealistic nonsense, the impact of trenchant analyses like this one will be net zero.